192815 research outputs found
Sort by
LLM-CompDroid: repairing configuration compatibility bugs in android apps with pre-trained large language models
No full textXML configurations are integral to the Android development framework, particularly in the realm of UI display. However, these configurations can introduce compatibility issues (bugs), resulting in divergent visual outcomes and system crashes across various Android API versions (levels). In this study, we systematically investigate LLM-based approaches for detecting and repairing configuration compatibility bugs. Our findings highlight certain limitations of LLMs in effectively identifying and resolving these bugs, while also revealing their potential in addressing complex, hard-to-repair issues that traditional tools struggle with. Leveraging these insights, we introduce the LLM-CompDroid framework, which combines the strengths of LLMs and traditional tools for bug resolution. Our experimental results demonstrate a significant enhancement in bug resolution performance by LLM-CompDroid, with LLM-CompDroid-GPT-3.5 and LLM-CompDroid-GPT-4 surpassing the state-of-the-art tool, ConfFix, by at least 9.8% and 10.4% in both Correct and Correct@k metrics, respectively. In addition, our real-world evaluation shows that LLM-CompDroid successfully repairs 21 configuration compatibility bugs with a 100% success rate, demonstrating its practical utility. This innovative approach holds promise for advancing the reliability and robustness of Android applications, making a valuable contribution to the field of software development
An uneven playing field: a mixed methods, multiphase feasibility study of a programme to reduce gambling among at-risk men in a professional football club setting
Get PDFBackground:
Sports betting is a growth area for the gambling industry, with football fans a key target of advertising. Men are particularly at risk from gambling harm. The 8-week Football Fans and Betting (FFAB) intervention was designed for delivery in professional football clubs by club community coaches to reduce sports betting and other forms of gambling among men aged 18–55 with a PGSI score of < 15. This paper reports the acceptability and feasibility of delivering FFAB in England.
Methods:
We conducted a multiphase, mixed methods, feasibility study of the FFAB intervention, assessing feasibility through three criteria: recruitment, fidelity and acceptability. We generated quantitative process and attendance data to assess recruitment and retention, and observation, interview and focus group data to examine fidelity and acceptability. Quantitative data were analysed descriptively, qualitative data using thematic analysis, before being combined using triangulation protocol to assess feasibility against the three criteria.
Results:
FFAB was launched in six clubs. Despite multiple attempts to refine and improve recruitment processes, no club was able to recruit the target number of participants to the programme. Retention on the programme also faced challenges. The saturation of football’s commercial landscape by the gambling industry and stigma-related social dynamics appear to underpin these findings. Due to recruitment and retention challenges, fidelity criteria were not met. However, participants that attended the programme, and coaches that delivered it, reported that FFAB was acceptable and supported some to make changes to their gambling behaviours. Throughout the qualitative dataset, participants and coaches emphasised the necessity of intervention to prevent gambling harms among men.
Conclusions:
We found that FFAB was acceptable to the coaches who delivered it and the participants who attended. However, our model for recruitment did not work, with consequences for fidelity. We also faced difficulties with retention. More feasibility work to develop a different approach to a gambling reduction with men between 18 and 55 with a PGSI score of less than 15 is required
Visual perception-inspired 3D point cloud sampling
Get PDFTask-oriented sampling aims to predict the importance of points of a point cloud to better serve downstream tasks, which has attracted increasing attention in the fields of computer vision and visualization in recent years. However, existing methods cannot sufficiently leverage both global saliency and local saliency cues, resulting in suboptimal performance that requires further improvement. To tackle this challenge, we propose a novel 3D point cloud sampling method inspired by the human visual perception mechanism in this study, which can effectively extract important point cloud subsets from critical regions to better adapt to downstream tasks, thereby maintaining superior sampling performance. The proposed Visual Perception-inspired 3D Point Cloud Sampling (VPI-3DPS) method simulates the human visual system’s dynamic attention-shifting strategy by combining coarse-grained attention-driven sampling with fine-grained detail preservation. This allows our approach to adaptively capture both global context and local details within point cloud data, safeguarding downstream task performance. By leveraging Gated Recurrent Units (GRUs) for long-term dependency modeling and integrating Graph Convolutional Networks (GCNs) to capture local structures, VPI-3DPS obtains an integrated representation of regional correlation and detail awareness. Extensive experiments show that VPI-3DPS outperforms existing methods. Compared to the best-performing approaches, it achieves an average increase of 1.29% in classification accuracy, an average reduction of 13.20% in registration MRE, and an average decrease of 4.29% in Chamfer Distance for reconstruction
AI & The Paradox of Agency
No full textInternational group exhibition including new commissions and accompanying catalogue curated by Cook, Bildmuseet, Umea, Sweden, opening March 13 2026. Supported by Umeå University, UmArts, WASP-HS and the Jacob Wallenberg Fun
Developing research resources and minimum data set for care homes' adoption and use
No full textBackground:
In England, care homes are the primary providers of long-term care for older adults. The increasing recognition of the importance of social care underscores the importance of collaboration between the National Health Service and care homes. The lack of data sharing among stakeholders limits opportunities for co-ordinated care, service development and research.
Objectives:
1. Identify how to support research, service development and innovation in care homes.
2. Combine existing evidence with care home-generated resident data to create a minimum data set that is relevant and usable for stakeholders, including residents, relatives, practitioners, researchers, regulators and commissioners.
Design and methods:
The study used a mixed-methods approach, structured into five work packages, supported by patient and public involvement and engagement with residents, carers and staff:
- Work package 1: Conducted two evidence reviews on outcome measures and factors enhancing research productivity in care homes.
- Work package 2: Created a trial archive for secondary data analysis.
- Work package 3: Conducted a scoping review, a realist review and a national survey to define minimum data set content and assess implementation challenges in English care homes.
- Work package 4: Linked residents’ data from National Health Service and social care data sets with data from study care homes, deriving useful minimum data set variables and assessing data quality.
- Work package 5: Piloted the minimum data set at two points in care homes within three integrated care systems, conducted focus groups and interviews with care home and integrated care system staff.
Three national consultations explored how stakeholders use resident information, measure quality of life and minimum data set usefulness. Additionally, subprojects examined data availability in domiciliary settings, staff reasoning when assessing resident well-being and completing research during rapid policy changes.
Findings:
- The reviews revealed significant heterogeneity in outcome measurement and questioned the appropriateness of some methods and measures used for care home research.
- The Virtual International Care Home Trials Archive merged data from 6 United Kingdom randomised controlled trials with 5674 residents across 308 care homes.
- International minimum data set studies are a valuable resource for international comparative research. The wide range of measures used are mostly clinical with under-representation of measures important to care homes (e.g. quality of life).
- A national survey of care homes demonstrated the range of information, including clinical measures being routinely collected.
- The realist review identified motivation, front-line staff monitoring and embedded recording systems as important for minimum data set implementation.
- The pilot study recruited 996 residents from 45 care homes, with 727 residents’ data included in the minimum data set. Residents’ digital care records were linked to statutory health and social care data sets, creating a viable minimum data set prototype with metadata as resource.
Conclusions:
The study provided an evidence-based critique of care home research and a resource for secondary data analysis for future research. It developed a prototype minimum data set linking National Health Service, social care and care home data, demonstrating its importance as a basis for discussions between health and care staff.
Limitations:
The COVID-19 pandemic disrupted relationships and recruitment. Governance challenges prevented linking residents’ data to general practitioner records.
Future work:
Future research should assess whether the care home minimum data set improves resident outcomes, service delivery, staff experience, cross-sector collaboration, resource use and digital technology implementation
Effects of long-term wetland variations on flood risk assessments in the Yangtze River Basin
Get PDFFlooding is the most frequent natural disaster in the Yangtze River Basin (YRB), causing significant socio-economic damages. In recent decades, abundant wetland resources in the YRB have experienced substantial changes and played a significant role in strengthening the hydrological resilience to flood risks. However, wetland-related approaches remain underdeveloped for mitigating flood risks in the YRB due to the lack of considering long-term wetland effects in the flood risk assessment. Therefore, this study develops an wetland-related GIS-based spatial multi-index flood risk assessment model by incorporating the effects of wetland variations, to investigate the long-term implications of wetland variations on flood risks, to identify dominant flood risk indicators under wetland effects, and to provide wetland-related flood risk management suggestions. These findings indicate that wetlands in the Taihu Lake Basin, Wanjiang Plain, Poyang Lake Basin, and Dongting and Honghu Lake Basin could enhance flood control capacity and reduce flood risks in most years between 1985 and 2021 except years with extreme flood disasters. Wetlands in the Sichuan Basin have aggravated but limited impacts on flood risks. Precipitation in the Taihu Lake Basin and Poyang Lake Basin, runoff and vegetation cover in the Wanjiang Plain, GDP in the Taihu Lake Basin, population density in the Taihu lake Basin, Dongting and Honghu Lake Basin, and the Sichuan Basin are dominant flood risk indicators under wetland effects. Reasonably managing wetlands, maximizing stormwater storage capacity, increasing vegetation coverage in urbanized and precipitated regions are feasible suggestions for developing wetland-related flood resilience strategies in the YRB
Beyond questions: leveraging ColBERT for keyphrase search
Get PDFWhile question-like queries are gaining popularity, keyphrase search is still the cornerstone of web search and other specialised domains such as academic and professional search. However, current dense retrieval models often fail with keyphrase-like queries, primarily because they are mostly trained on question-like ones. This paper introduces a novel model that employs the ColBERT architecture to enhance document ranking for keyphrase queries. For that, given the lack of large keyphrase-based retrieval datasets, we first explore how Large Language Models can convert question-like queries into keyphrase format. Then, using those keyphrases, we train a keyphrase-based ColBERT ranker (ColBERTKP QD) to improve the performance when working with keyphrase queries. Furthermore, to make the model more flexible, allowing the use of both the question and keyphrase encoders depending on the query type, we investigate the feasibility of training only a keyphrase query encoder while keeping the document encoder weights static (ColBERTKP Q). We assess our proposals’ ranking performance using both automatically generated and manually annotated keyphrases. Our results reveal the potential of the late interaction architecture when working under the keyphrase search scenario. This study’s code and generated resources are available at https://github.com/JorgeGabin/ColBERTKP
Enhanced reversible barocaloric effect at low pressure in neopentyl plastic crystal solid solutions
Get PDFThe discovery of colossal barocaloric effects in neopentyl glycol (NPG) makes plastic crystals promising candidates for solid-state refrigerants with lower environmental impact than vapour compression fluids. Optimising operational temperatures and low-pressure operability remains challenging without compromising thermodynamic parameters. Here, we implement a strategy to improve the viability of NPG derivatives as barocaloric refrigerants. We blend pentaglycerine (PG) with NPG to lower the phase transition temperature, then dope the blend with 2% pentaerythritol (PE) to improve transition reversibility. In comparison with NPG under the same conditions, this ternary system has a seven-fold increase in reversible isothermal entropy change (|ΔSit,rev| = 13.4 J kg-1 K-1) and twenty-fold increase in operational temperature span (ΔTspan = 18 K) at pressures of 1 kbar. Synchrotron x-ray diffraction and quasielastic neutron scattering reveal structural and dynamical effects that broaden the temperature range of the first-order phase transition due to intermolecular hydrogen bond network disruption by the molecular dopants. We propose that exploiting the compositional phase space of multi-component molecular blends is effective for designing practicable molecular BCs
Data sources:
Access Repository Dashboard
Do you manage Open Research Online? Become a CORE Member to access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard! 👇